Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Episode #41: Mike Pell
142 Plays9 years ago
As everyone knows, Microsoft creates a lot of stuff: software, hardware, services, cloud services, and more. But there is also a group at Microsoft that gets to play around and create cool stuff. In this week’s episode of The PolicyViz...
The post Episode #41: Mike Pell appeared first on PolicyViz.
Recommended
Transcript
Introduction and Sponsor Highlight
00:00:00
Speaker
This episode of the PolicyViz podcast is brought to you by Juice Analytics. Juice is the company behind Juicebox, a new kind of platform for visualizing data. Juicebox is a platform designed to deliver easy to read interactive data applications and dashboards. Juicebox turns your valuable analyses into a story for everyday decision makers. For more information on Juicebox or to schedule a demo, visit juiceanalytics.com.
Meet Mike Pell: Designer at Microsoft Garage
00:00:32
Speaker
Welcome back to the Policy Vis podcast. I'm your host, John Schwabisch. I'm excited to welcome to the show this week, Mike Pell from Microsoft. Mike is a designer and envisioner, which is kind of an awesome title at Microsoft in a group called The Garage. Mike, welcome to the show. Hey, John, thanks for having me.
00:00:49
Speaker
Great to have you on. I want to start by maybe having you introduce yourself and talk a little bit about your background. And then also maybe if you can talk a little bit about what you do in the garage at Microsoft, which in addition to having an awesome personal title is an awesome name for a group to work in. Just sounds like you're hanging around with friends building cool stuff.
00:01:07
Speaker
Let's start there. That's exactly what we do, John. That way. First of all, thanks for having me on. The garage is a really tiny team, but it represents a giant community around Microsoft worldwide where we help employees and small teams really follow their passions, whether it has anything to do with what they do for their day job or not. We're really trying to help people get their best ideas out in front of real people who can help them develop the idea, iterate,
00:01:35
Speaker
We collect data we help people ship their ideas through microsoft dot com like garage and we also sort of do lots of internal hackathons and we're helping to change the culture as much as we can i'm sure you've been reading a lot about how microsoft has changed in the last two years since i don't know it took over a ceo and i think you know the garage has quite a bit to do with some of the cultural change.
Mike Pell's Career Journey
00:01:58
Speaker
And you've been in this space of information visualization, computer graphics for a while. Can you talk a little bit about sort of quickly what your career trajectory has been? Absolutely. So it's interesting to note that I started off as an artist. And when I was going to school for art, I actually started programming. And by accident, I thought, hey, I can make more money doing this programming thing.
00:02:18
Speaker
But I've always been able to do both the design side and the analytics and the coding part. So I started off way back at the dawn of the Macintosh. I started one of the first Macintosh companies called beyond with a friend Stuart Davidson. And we learned everything you can learn about running a software company. That was the good old days. But I've always been involved in.
00:02:39
Speaker
things that were really more information-based and visual. And so, you know, over the years, I've worked for companies like Adobe, lots of Silicon Valley startups, and now Microsoft for the last 15 years. And all along the way, have been very, very fortunate to be able to stay in this, you know, design and coding both types of jobs. In the garage, a lot of the things that I do are help teams when they come in
00:03:03
Speaker
Just sort of assess what type of experience you know they're trying to provide for people whether it's You know that they're putting out a maker recipe through an open source or they're putting out an Android or iOS app You know or even just creating a new web service like some of the new ones that we put out like fetch And where you can I don't know if you saw fetch is very funny
00:03:23
Speaker
You can feed a picture of your dog, and it'll tell you what breed the dog is, but if you put in the picture of a person, it'll tell you what kind of dog the person is. That's the kind of stuff that we do in the garage. It's very fun. So you were right. We're having a blast. Nice,
Understanding Data through Visualization
00:03:38
Speaker
nice.
00:03:38
Speaker
So you've been doing this a long time and you've sort of seen the transition as it were from different types of interfaces. And I know you've also been working on this idea of how do we get people to sort of walk away or look at a visualization look at some data and sort of get that you know that bottom line or understand the story right away.
00:03:57
Speaker
I think what we were talking about earlier, what you sort of call the moment of clarity. So this is a really interesting topic. I think we should kind of spend some time talking about it. So how do you sort of define the moment of clarity and how does it differ across the different platforms and ways in which we visualize and communicate data? Yeah, specifically in visualization.
00:04:14
Speaker
There's a point where we can put together any type of visualization on the spectrum, whether it's something very factual or something that's even data art. But there's a moment where the person who's consuming that piece will either get it, hopefully, as the author intended, or they'll sort of struggle with what was meant by it. And so in that time that you're looking at a piece, whatever it may be,
00:04:40
Speaker
there's a particular moment where most people will say they got it. You know, it's the aha moment. And as a designer, you know, and yourself, of course, you know, somebody who does a lot of presentations and talks, you have to try your best to get your point to be very clear. I've always, you know, over my career, I've tried to practice what I call a radical simplification, right? It's getting rid of a lot of extraneous things and also focusing on the clarity aspects.
00:05:08
Speaker
But the fact is, you know, people are very, very different. The situation is different. What emotional state are you in when you're looking at this? What's your background? You know, like how deep have you been exposed to any of the information before? So the moment of clarity varies from person to person situation to situation.
00:05:25
Speaker
But I've always had this hypothesis that we could do a much better job of engineering the moment of clarity as, you know, data scientists and designers and coders, you know, and people in this in this community, whereas I think it's taken off, you know, sort of one off sort of away right now. So I think there's a lot of room that we can improve. Yeah. So how do you view the different
00:05:49
Speaker
ways in which we communicate data from sort of your one off graph may or may not be static to sort of a big interactive data tool. I mean, how do those how do you how do you get people to get to that moment of clarity when you have such different ways to communicate data communicate information?
Methods of Data Presentation
00:06:08
Speaker
That's an excellent question. That's exactly the question that I asked myself and started to. I've been collaborating over the last six months with a guy from the London Business School named Mateo Visiton.
00:06:19
Speaker
And so Mateo and I, Mateo's a visual designer, but he's also a behavioral scientist. And so we've been exploring ways to measure and test how people react to the way that data is presented in certain ways. And I don't want to try to make this into something that is a template, because I don't think you can, frankly.
00:06:41
Speaker
But I do think that we can put together a set of principles or guidance that can help the authors, the individuals and the teams putting information together or pieces together to be able to help people get to the moment of clarity, perhaps more quickly or efficiently than they have in the past. A lot of times, there are so many different types of visualizations. There's a very stock
00:07:05
Speaker
you know, charting and things that you would find inside of Tableau or Power BI or whatever. And there's also pieces that are much more thoughtful, you know, like you had Kim Rees on a couple of weeks ago. And, you know, certainly the pieces that Periscopic does are quite thoughtful and they take a little bit to sort of digest, you know, in some ways. And then there's other pieces that are just purely meant as art and they're based on real data. But, you know, but they really might take some time to soak in to really understand
00:07:34
Speaker
So for any of those things, the methods that we've discovered that you can use really vary from person to person to the point where we feel like we have to offer multiple ways or multiple dimensions that you can consume this information.
00:07:51
Speaker
And so that's sort of been the key for me is that there is no such thing as one size fits all as far as moment of clarity goes. So we as, you know, tool builders and platform builders and artists and designers and data scientists, we can do a much better job of presenting information in ways that allow the person to go in, you know, at the altitude that they're comfortable with, whether it's high level summary, low level detail, you know, or how broadly or how narrow they choose to consume it.
00:08:21
Speaker
I think that last point is probably part of the key because in some cases you are putting out a visualization that is for a specific audience. Think of an academic journal. You write your academic journal article, you know that the reader is sophisticated.
00:08:40
Speaker
That's a sophisticated reader who knows the field that is reading the article for that reason. But then you have other pieces that you're trying to hit multiple audiences, maybe trying to hit a broader audience. And so how do you balance trying to sort of meet the needs of all these different audiences with different interests and different skills and different sophistication? How do you sort of get that moment of clarity for different kinds of user?
Design Challenges for Diverse Audiences
00:09:03
Speaker
Exactly. I think you'll probably just have the answer for this, right? I kind of think I do. In the long run, the tools and platforms that we use will have the ability to allow you to construct
00:09:17
Speaker
pieces, I'll just call them pieces of content, whether they're charts or narratives or PSAs or whatever, that allow the reader, the consumer, to view them and experience them at what I call altitudes, at different altitudes. And as the author, you may pick one of those altitudes, let's say, high-level summary to be the default, but there's enough data to allow people to drill in. There's enough cross-references and linkages to allow people to explore the periphery and related topics
00:09:47
Speaker
And so I think that in the long run the tools and platforms are really the things that will allow us to build what I call smart information is no pieces that are very fluid and much more malleable in the way that they can be consumed.
00:10:01
Speaker
And in the short term, I think it really comes down, unfortunately, to the authors. We have to stop producing dead pixels. You know what I mean? We have to stop making things that can't be explored and prodded and drilled into. And I'm not saying you have to ship all of the capabilities of Tableau or something with each piece.
00:10:23
Speaker
but you need to have the ability to consume it in the way that's right for you. And that's unfortunately in the short term going to be on the shoulders of the authors of these things.
00:10:33
Speaker
I'm curious, there's for a while been, I'll call them interactive visualizations, but they're basically static visualizations that have some sort of interactivity just sort of, you know, layered on top. So it's a line chart, there's three lines on it. And it allows you, you know, the visualization allows you to click a button or click a point or click a line and it highlights. And I wonder whether you think that level of interaction
00:10:56
Speaker
actually helps people get this moment of clarity or if it's just sort of gratuitous and going forward, whether you think that type of interactivity is not going to be as common is not going to be as you so much because it doesn't necessarily help people better understand just allows them really just to play with something.
00:11:12
Speaker
Yeah, you're absolutely right. I'm pretty against what I call putting different dimensions on rails. Just isolating a few variables and letting people move some sliders around. That's fine if that's going to help you to get a little bit more of the detail in a piece. But really it comes back to the design.
00:11:32
Speaker
This is all about people right it's about you know what did the author know the team what did they really think the outcome of this was gonna be like what were they trying to get across were they trying to be completely factual like let's take for example. New York Times is gonna present you know an election result.
00:11:49
Speaker
they would like to be as factual as humanly possible, right? And the presentation of that information versus let's take something like, you know, back to periscopic, where if you look at their gun piece, that to me, that's, you know, sort of like a PSA, there's a message there, intentional, right? You know, they're not, you know, hitting you over the head with it, but it's pretty clear what their point of view was.
00:12:15
Speaker
And so as the authors, as the creators of information, we really need to focus on what we're trying to get across. And I'm very fond, as you know, of saying, just tell me. When you're designing a visualization, just tell me. Don't make me sit there for minutes or hours trying to figure out what you're trying to tell me through the way that you've chosen to depict something. And oftentimes,
00:12:41
Speaker
we have been skewing towards beauty or visual interest rather than being clear. And being clear is hard. You know this, you have to present all the time. It's very hard to be clear and concise in a short amount of time, but yet you sort of have to isolate that. You need to figure out what is it that is the most valuable thing to get across to people and how can you do that without ruining the aesthetic or the artistic integrity of a piece that you're working on.
00:13:10
Speaker
And where do you think that line is between sort of the clarity of presenting information and design? I'll give you an example. I was rereading a Medium post earlier today actually by Martin Wattenberg and Fernando Villegas. I think that was early last year, middle of last year about critique in the data visualization field. And one of the examples they showed
00:13:30
Speaker
was sort of a radial chart of things going on in the Middle East. And so you had this sort of, you know, 10 or so countries going around this circle and sort of a big radial chart. And Alberto Cairo from Miami had redesigned it sort of more of a, it was basically a dot plot or a column chart. So it sort of got rid of the circles part, right? Because, you know, his argument was, well, you don't really know. Circles are evil, right? Circles are evil, right. But, you know, and you see these sorts of things all the time. Let's get rid of the circles and let's use the bar charts.
00:13:56
Speaker
But sometimes the circles, there's an aesthetic beauty to it or something that's attractive to it, but not necessarily allowing you to sort of dig in and see all the detail and all minutia of the data. So I'm curious, how do you think about the balance between the aesthetic that pulls the reader in versus a sort of, you know, maybe more of a, this is a lot clearer because I have bars instead of circles.
00:14:19
Speaker
Yeah, well, you know, I very often times I talk about being beautifully clear. And the point there, and I completely agree with Alberto most of the time about trying to get back to the, I would say, the most clean and accurate representation of data as you can, but you have to have some form of aesthetic or visual interest or people just blow by it, right? I mean, like, why would you even bother to stop to look at another line chart? Like, who cares?
00:14:49
Speaker
And so there's a lot of room to introduce through callouts, through the way it's laid out, through sort of helping people focus on the important parts of the charts. Like I've come across something in Unflowing Data, you know, Nathan was talking about this taxi ride volume during the Super Bowl chart that had come out right in Upshot. It was super interesting. There was a great example of helping people focus on an important piece of information. So right in the middle of this, you know, thing that's sort of like a histogram looking chart,
00:15:17
Speaker
It says Katy Perry halftime show this is back when the Seahawks and the Patriots were playing the Super Bowl last year. So at halftime you see that there are not a lot of taxi rides going on because it's like very exciting halftime with Katy Perry for watching.
00:15:31
Speaker
But the thing that was very subtle in this was at the very end of the games, you know, all Seattle Seahawks fans remember very horribly. There was an interception to end the game. And all of a sudden, there's this giant spike, you know, in taxi rides right after that, because everyone's like, Oh, right. Right. But you know what? In the chart, I didn't remember
00:15:50
Speaker
because this chart looked fairly normal. But wow, that's super interesting to me as a Seahawks fan, right? Football fan. But there was no call-out about that. There's just a little thin gray line. There was an opportunity there for the author to help me focus on the important parts of the data without taking a point of view, without having an opinion about it. I think we miss a lot of opportunities.
00:16:18
Speaker
to point out the interesting parts of things. Maybe it's not with circles.
00:16:23
Speaker
But the fact is, as designers, as data scientists, as people trying to get this information across, we can't be afraid to point out what our insights tell us. And a lot of times, people shy away from that. They're trying to be too factual. They're afraid to... They'll call it embellishing. It's not embellishing. It's actually clarifying, if you ask me.
00:16:48
Speaker
Right, right, right. So it's not just that you don't let the sort of data speak on their own, you help it out a little bit, you help the reader out a little bit. Yeah, because in this day and age, I mean, who has time to sit there and try to, you know, understand, like try to dissect, try to, you know, because it may not be your sort of field, right, of expertise. And you may just be looking at something casually, but yet the author would really like you to come away with something. And so very often we miss that opportunity.
00:17:15
Speaker
It's interesting the way you talk about communicating data and communicating information because you have this emphasis of talking about the user being a
Emotion and Storytelling in Data
00:17:24
Speaker
person. The reader's a person. The person who's created it is a person. And so it sort of calls back to an earlier episode of the podcast with Kim Reese and Mushon where we were talking about empathy and data visualization.
00:17:36
Speaker
and so i wonder what you think about trying to get readers to really connect with the visualization connect with the data and feel something you know and i think there are different kinds of visualizations that get us to feel um you know maybe the taxi cab example we feel something because we all watch the super bowl and a visualization on you know some microbiology thing that only four people in the world understand is a different type of thing but
00:17:59
Speaker
What is the impact or what is the effect of drawing out emotion and drawing out empathy? How important is that when people are creating visualizations and trying to tell stories with their data? Oh, hugely important.
00:18:12
Speaker
Absolutely, hugely important. I mean, as a designer, I'm all about people and emotion and understanding what people are thinking and feeling. And trying to use that, I think Kim even mentioned that a few weeks ago that she's not even as a data scientist, she's not opposed to trying to draw on emotion to get people engaged. It is hugely important. And I am also a big fan of persuasion.
00:18:35
Speaker
And a lot of people would say, oh, you know, well, there's no room for persuasion in data visualization. I would say that's nonsense. You know, it is, you know, people like to call it data storytelling sometimes. Right. But the fact is.
00:18:47
Speaker
we should employ all of those methods and all of those tools and all of those techniques that the great storytellers and playwrights and people who make movies have used, you know, for so many decades to try to get whatever that is that's important, you know, as an author, whatever I'm trying to get across or whatever my team is trying to get across, let's use everything at our disposal. I mean, as designers, you know, we're not above lying, cheating and stealing, you know, to get our way, right?
00:19:15
Speaker
And I think the same holds true and in any type of information, communication, information visualization or data visualization for that matter, we need to draw on all of that talent and all the techniques we have to get our points across. And I don't think you're going to bias somebody so much.
00:19:34
Speaker
unless you actually try to. If you're trying to be completely persuasive and you're trying to really overtly twist the data to support your point of view, that's one thing. If that's what you wanted to do, then great, go do it. But if you're just trying to present information clearly, I do think there's a big need to draw on the people part of it, on the emotional part.
00:19:59
Speaker
You touched on a couple of things there. Persuasion a lot of times has this negative connotation to it. Absolutely. Visualizations are almost always trying to persuade people that the thing that they're showing is true. The unemployment rate is going down. This thing is going up. There's all levels of persuasion. I guess the difference is whether you're being honest with the data and trying to persuade someone that your perspective is correct, given that you are using the data honestly.
00:20:28
Speaker
And then the situation where you're distorting the data either visually or in the way you've created the final product with the data itself and you've distorted it in different ways. I'm also curious when it comes to persuasion when it comes to getting people to connect sort of emotionally to visualization.
00:20:45
Speaker
how important it is to connect at the individual level. So we've mentioned the periscopic guns piece. And one thing about the guns visualization is that it has a line for each individual in their data set. How important is it for a creator to convey those individual stories and maybe not sort of always aggregate into six bars, but show more of the individual observations, the individual stories.
00:21:13
Speaker
Yeah, well, again, you know, it's always context dependent, right? Sure, sure. Whenever possible, I would always tend to try to go to the individual person if possible, whether it's in the data or whether it's in the consumption of that. So maybe I'm talking about a very big concept, like, you know, like, let's talk about very difficult world issues.
00:21:33
Speaker
If I can't make a personal connection with what you're telling me, then chances are I'm going to move on to something else. So super important for you, the author, to try to get me somehow as an individual interested in what you're talking about. So I can dig in. So I can actually, you know, get all the benefits of what you're trying to do. In some cases, I don't think that's possible. I just don't think that you can involve an individual person as well as, you know, as the periscopic piece did.
00:22:02
Speaker
There's a certain point where you do have to sort of abstract things. That doesn't mean that you can't use other elements. Not every data visualization has to be on a plain white background. There's a lot of things we can do from the design of it, from sound and motion and the way that it's staged. All the cinematic elements come into play to sort of make
00:22:27
Speaker
You know a like even you know going back to that piece as an example very beginning of that piece is very cinematic right background yeah 2010 Something just happening very slowly. You know all of a sudden the other data starts to draw itself. It's it's like a story unfolding right yeah, and all of a sudden I get the moment of clarity because at about the sixth person it starts to accelerate and it hits me oh and
00:22:50
Speaker
Wow, they're talking about a ton of people dying and so much tragedy and, you know, like loss of years of life. I mean, it sort of gets you, but it gets you because it was done in a cinematic way. So perhaps even the most mundane, straight up,
00:23:10
Speaker
you know, I would say, you know, data pure type of representation that you could do could actually be presented to help make an individual connection with me to consumer or if it's representing people to sort of show, you know, like we were talking about sand dance earlier, every single pixel on the screen represents a unique piece of data, which may be a person, right?
00:23:32
Speaker
Yeah, and so for those who don't know, Sand Dance is the new project coming out from Microsoft Research, where Mike's garage is sort of a part of.
Sand Dance: A New Tool for Data Exploration
00:23:40
Speaker
You want to talk about real quickly before we wrap up for people who haven't seen it, haven't heard of it yet? Sure. Sand Dance is a really great piece of visualization work done by Stephen Drucker and Roland Fernandez in Microsoft Research that you'll be able to play around with before too long.
00:23:55
Speaker
And the premise of this is that they've been working on this for years and you can go find it, but they really wanted to help people explore data. So to me, it's a great example of a new genre of tool at our disposal called a data explorer.
00:24:09
Speaker
And it's super fluid in order to move through data sets. You can load your own data and play around with data sets that are already there. But it allows you to look at things very quickly, move on, sort of explore, play what if, try different things out super fast in a way that we've never been able to do before.
00:24:27
Speaker
And they've constructed this tool in a way that it can be extended, and they're going to continue to iterate on it and experiment. And it's really about helping the visualization community try out new ideas and hopefully tell better stories with data using that particular tool. Great. Well, on that note of new tools to explore and trying to maybe be a little more aesthetic and beautiful in our visualizations, even with the boring stuff, Mike, I want to thank you for coming on the show. This has been really interesting. Thank you, John.
00:24:57
Speaker
And thanks to everyone for listening. If you have any comments or suggestions, please let me know and please rate the show on iTunes or your favorite podcast provider. It helps other people learn and know about the show. So until next time, this has been the Policy Viz Podcast. Thanks so much for listening.
00:25:24
Speaker
Again, thanks to our sponsors, Juice Analytics. For 10 years, they've been helping clients like Aetna, the Virginia Chamber of Commerce, Notre Dame University, and U.S. News & World Report create beautiful, easy to understand visualizations. Be sure to learn more about Juicebox, a new kind of platform for visualizing data at juiceanalytics.com. Also, check out their book, Data Fluency, found on Amazon.